Today, the approaches available for the distribution and summarization of network connected device capability information in a network are limited to a centralized model and rely on the use of computational power to provide matching of contextual types and tags to items of interest. As the number of advertisements grows, the number of potential matches of queries to advertisements grows. Some matches may have nothing to do with the topic of interest of the entity making the query, thereby creating a larger proportion of uninteresting responses to queries as the size of the query space grows. Also, with a centralized model, the collection of contextual types and tags has to be done prior to receiving and processing queries to search the database. Therefore, the database will always be outdated.